Beep6581 / RawTherapee

A powerful cross-platform raw photo processing program
https://rawtherapee.com
GNU General Public License v3.0
2.82k stars 319 forks source link

Fuji X-T2 Support - White Frames #5824

Closed dancat closed 4 years ago

dancat commented 4 years ago

Hi,

I took some sets of white frames for Fuji X-T2 as specified in http://rawpedia.rawtherapee.com/Adding_Support_for_New_Raw_Formats#White_Levels

They can be found in https://filebin.net/pd90j4akospyzca4

Let me know if more are needed.

All the best, Daniel

Thanatomanic commented 4 years ago

Thanks!

Thanatomanic commented 4 years ago

Interestingly, the sensor seems to be slightly non-uniform. Take a look at this: image

This is for DSCF7302.RAF (no LNR). Values scaled to enhance the differences. The actual difference in 14-bit raw values is a mere ~500 away from 2^14.

dancat commented 4 years ago

Interesting indeed ...

I read on dpreview some analysis regarding the use of the X-T2 for astrophotography and their finding was that the camera applies some sort of filtering over the raw data, what you see might be caused by this?!? I think the effect was very visible with longer exposures (over 5 seconds)

Another thing that might matter, for this type of cameras is the type of shutter used. In the same analize they found that using the electronic shutter or the continuous shooting mode makes the camera to switch to 10 bit. Should I make for this case another set of whites using the electronic shutter?

Thanatomanic commented 4 years ago

Possibly related and similar to what you say about what you read on dpreview: https://discuss.pixls.us/t/handling-fujifilm-x-trans-auto-focus-pixels/19047

I'm not sure how to approach this properly...

dancat commented 4 years ago

I have also took the same frames for the X-T3 and will upload them soon. However, I might have the LENR On files for X-T2 wrong, as the LENR was not applied to all of them. Should I redo them, making sure it does apply?

Thanatomanic commented 4 years ago

@dancat If you think that you need to redo some shots, please provide a full new set.

If you're going to redo things, please also pay attention that some low ISO shots were not fully overexposed in all channels (particularly DSCF7290.RAF). You can easily set the shutter speed to 1/10s to compensate. Also, according to the Exif metadata DSCF7309.RAF is ISO12800, DSCF7310.RAF is ISO 200, and DSCF7330.RAF is ISO12800 although your notes mention they should have been ISO25600, ISO100 and ISO25600 respectively. Can you double check that your camera is setting those values properly in the Exif?

dancat commented 4 years ago

I'll try to be more careful :)

I am using the camera overexposure highlights and take the photos when all the screen blinks. But not being a raw histogram it can be that it is not overexposed completely. Is there a better way to do it?

Thanatomanic commented 4 years ago

Not really, just be very overexposed. If you do what you did initially, but set shutter speed to 1/10s or even 1/2s you're nearly 100% sure you've overexposed your image.

dancat commented 4 years ago

OK, In this case I will redo the sets and check also what happens with the extended ISO values.

heckflosse commented 4 years ago

@Thanatomanic

Interestingly, the sensor seems to be slightly non-uniform.

Just a dumb thought: Could that been caused by reflecting light from sensor back to rear element of lens and then back to sensor ?

dancat commented 4 years ago

I've put the new files here: https://filebin.net/h4441zat3nyf5maa I took care that:

Should I redo also the Set3 for aperture values?

dancat commented 4 years ago

BTW, I checked also the extend ISO range and the camera is writing it wrong in the EXIF. In Set1 and Set2 the first and last picture are the extended LOW (100 but camera writes 200) and extended High (25600 but the camera writes 12800).

The Fuji X-T3 writes the correct values. I will upload soon also the files for this camera.

Thanatomanic commented 4 years ago

@dancat Thanks for reuploading and clarifying the ISO values. I've downloaded the images and will inspect them soon. Please upload the X-T3 files in a separate issue, just to keep it ordered :-)

Thanatomanic commented 4 years ago

@dancat Similar to your X-T3 samples I determined some useful white levels. Can you test them by finding and replacing the X-T2 data in your camconst.json and looking at a few regular and overexposed photos?

{ // Quality B
        "make_model": [ "FUJIFILM X-T20", "FUJIFILM X-E3", "FUJIFILM X100F", "FUJIFILM X-PRO2", "FUJIFILM X-H1" ],
        "dcraw_matrix": [ 11434,-4948,-1210,-3746,12042,1903,-666,1479,5235 ], // DNG_v9.4 D65
        // "raw_crop": [ 0, 5, 6032, 4032 ], // full raw 6160,4032, Usable 6032,4032 - for lossless compressed files
        // "raw_crop": [ 0, 0, 6032, 4032 ], // full raw 6160,4032, Usable 6032,4032 - for uncompressed files
        "raw_crop": [ 0, 5, 6032, 4026 ], // full raw 6160,4032, Usable 6032,4026 - for uncompressed and lossless compressed files (but reduces height by 6 pixels)
        "ranges": { "white": 16100 }
    },

    { // Quality A, samples provided by Daniel Catalina #5824
        "make_model": "FUJIFILM X-T2",
        "dcraw_matrix": [ 11434,-4948,-1210,-3746,12042,1903,-666,1479,5235 ], // DNG_v9.4 D65
        "raw_crop": [ 0, 5, 6032, 4026 ], // full raw 6160,4032, Usable 6032,4026 - for uncompressed and lossless compressed files (but reduces height by 6 pixels)
        "ranges": {
            "white": [ 16195, 16270, 16195 ] // typical LENR off: 16195, 16270, 16195
        } // With LENR on and ISO4000+ starts to overestimate white level, more realistic would be 16090
        // negligible aperture scaling effect
    },
dancat commented 4 years ago

Sure, I will give it a go the next days and provide feedback.

dancat commented 4 years ago

I've tested it with some older photos and I did not see anything wrong. I also used the various highlight recovery options and they were fine.

Thanatomanic commented 4 years ago

Thank you, updated camconst.json, closing.